Passive sampling and ecohydrologic modeling to investigate pesticide surface water loading in the Zollner Creek watershed, Oregon, USA

In the U.S. Pacific Northwest and California contaminants entering surface water may harm Endangered Species Act (ESA) listed salmonid species and consequently there is ongoing concern regarding agricultural practices and resulting pesticide surface water loading may adversely impact salmonid species, their food web, and habitat. Characterizing pesticide exposure in surface water at the watershed scale and beyond is challenging due to uncertainty regarding pesticide use practices and sparse monitoring data. We report here a 2-year continuous deployment of passive sampling devices (PSDs) for monitoring of pesticides in surface water at the outflow of the Zollner Creek watershed located within the Willamette Basin, Oregon, USA. This watershed is predominately agricultural and within the geographic range of two ESA listed Pacific salmonid species. Grab and passive sampling monitoring data were used to evaluate the performance of a probabilistic application of the Soil and Water Assessment Tool (SWAT), a physically based process model which integrates institutional and local knowledge and expertise to investigate the relationship between land use practices and pesticide surface water loading at the watershed scale. SWAT estimated pesticide surface water concentrations for the pesticides chlorpyrifos and trifluralin followed temporal trend in PSD monitoring results and the 5th to 95th percentile range of estimated pesticide concentrations based on the probabilistic assessment encompassed 65-76% of the observed PSD concentrations. Evaluation of model estimates for metolachlor in surface water was challenged by insufficient publicly available grab sample monitoring data. A process to estimate pesticide surface water concentrations on biologically relevant time scales and comparison to screening level aquatic life benchmarks is presented. Additionally, model estimates were used to characterize the variance in surface water concentrations in this small hydrologically responsive watershed to determine grab sampling frequency adequate for model evaluation.